Percutaneous Coronary Intervention Versus Coronary Artery Bypass Grafting in CKD

What Does This Important Study Show? 

The SYNTAX trial was a randomized multicenter international clinical trial evaluating CABG and PCI in patients with left main or 3-vessel CAD.7 The “all-comers” study design encouraged consecutive enrollment of all eligible patients (those with stable angina, unstable angina, and asymptomatic or atypical chest pain with demonstrated myocardial ischemia) without previous coronary revascularization. The study protocol did not exclude patients with CKD. Local interventional cardiologists and cardiac surgeons prospectively evaluated eligible patients to ensure that equivalent coronary revascularization could be achieved with CABG or PCI using paclitaxel drug-eluting stents. Patients eligible for both strategies were randomly assigned; other patients were entered into the PCI or CABG registries. The primary end point, based on intention to treat and evaluated at 12 months, was the composite of all-cause death, stroke, myocardial infarction, or repeated revascularization. The noninferiority study design had a prespecified delta value of 6.6%. Thus, if the 1-sided 95% upper confidence limit for the difference in event rates was <6.6%, PCI with drug-eluting stents would be noninferior to CABG. Subgroup analyses were prospectively designed, stratifying patients by CAD severity (SYNTAX score).

Of 4,337 patients screened, 3,075 (70.9%) met the eligibility criteria. A total of 1,800 patients were randomly assigned, with 897 assigned to CABG and 903 assigned to PCI. The remaining 1,275 patients (1,077 treated with CABG and 198 treated with PCI) were enrolled in a parallel nested registry because only 1 treatment option was suitable. Mean age was 65 years, and 77% of patients were men. Left main CAD was present in 39% of patients; <3% had a left ventricular ejection fraction <30%. Although >4 vessels were treated in both study arms, complete revascularization was more likely in CABG than PCI patients (63.2% vs 56.7%; P = 0.005). Periprocedural rates of major adverse cardiac and cerebrovascular events were similar for CABG and PCI (0.9% vs 0.3%; P = 0.13). At 1 year, the primary end point occurred for 12.4% of CABG and 17.8% of PCI patients (P = 0.002). The absolute difference in event rates was 5.5%, with an upper 95% confidence limit of 8.3%, exceeding the noninferiority margin. Thus, the noninferiority criterion was not met, indicating that PCI with drug-eluting stents was inferior to CABG. Similar results were found when the data were analyzed according to treatment provided (12.3% reached the primary end point in the CABG group vs 17.6% in the PCI group; P = 0.002). The 2 treatment arms did not differ significantly with regard to all-cause death or the composite of all-cause death, stroke, or myocardial infarction. Excessive repeated revascularization in PCI patients (13.5% vs 5.9% in the CABG arm; P < 0.001) was the primary driving factor. Interestingly, the stroke rate was higher in the CABG than PCI arm at 12 months (2.2% vs 0.6%; P = 0.003). In contrast, death rate from cardiac causes was lower in the CABG than PCI arm (2.1% vs 3.7%; P = 0.05).

Although the SYNTAX trial was not powered to evaluate the comparative efficacy of PCI and CABG in particular subsets (eg, patients with CKD), it prospectively stratified patients by CAD severity. Patients with low, intermediate, and high SYNTAX scores who were randomly assigned to CABG had similar 12-month rates of cardiac and cerebrovascular events (14.7%, 12.0%, and 10.9%, respectively). In contrast, the adverse-event rate in PCI patients was higher for patients with high SYNTAX scores (23.4%) compared with patients with intermediate (16.7%; P = 0.002) or low (13.6%; P = 0.04) scores.

How Does This Study Compare With Prior Studies? 

Randomized clinical trials evaluating outcomes for patients with left main and multivessel CAD date back to the 1970s.8, 9, 10, 11, 12 These landmark trials randomly assigned patients with stable angina to medical therapy versus CABG. High-risk subgroups with demonstrated improved survival with CABG included patients with left main stenosis >60%, 3-vessel CAD (stenoses >50% diameter), 2-vessel CAD with proximal left anterior descending stenosis ≥75% narrowing, and depressed left ventricular function. The invention of percutaneous transluminal coronary angioplasty in 1979 led to a plethora of randomized clinical trials comparing this technique with CABG.13, 14, 15, 16, 17, 18 Although these studies excluded patients with significant left main CAD, patients with multivessel CAD were enrolled. Several lessons were learned from meta-analyses and analytic reviews.19, 20, 21 First, in general, there was no difference in mortality or rate of myocardial infarction between the 2 strategies. The exception18 pertained to diabetic patients with multivessel CAD in whom surgical revascularization using an arterial conduit (left internal mammary artery graft) yielded improved survival compared with coronary angioplasty. Second, patients who underwent CABG were more likely to be free of angina at follow-up. Third, the rate of restenosis associated with early angioplasty techniques was >30%, resulting in higher coronary revascularization rates than observed in patients initially treated using CABG.22

Development of the Palmaz-Schatz bare-metal stent (Cordis,www.cordis.com) in 1994 led to reduced angiographic restenosis and, consequently, a reduced likelihood of repeated coronary revascularization.22 The ARTS-1 (Arterial Revascularization Therapies Study) and SoS (Surgery or Stenting) trials randomly assigned patients with multivessel CAD to bare-metal stent placement versus CABG.23, 24 The ARTS-1 trial found no significant difference between these revascularization strategies regarding mortality or the composite of death, myocardial infarction, and stroke.23 However, event-free survival was worse in patients randomly assigned to stent placement owing to higher rates of restenosis and repeated coronary revascularization. Outcomes with stent placement were worse for diabetic patients and patients with incomplete revascularization owing to higher revascularization rates. The SoS trial also found higher repeated revascularization rates in patients randomly assigned to bare-metal stent placement.24 In contrast to ARTS-1, the SoS trial reported higher mortality in patients with stents at 6 years. Meta-analyses of the clinical trials comparing both angioplasty and bare-metal stent placement with CABG have not shown a mortality difference during the first 5 years.25, 26 Subgroup analysis of patients undergoing stent placement documented a survival benefit for CABG emerging after the fifth year.20

The introduction of drug-eluting stents has resulted in reduced angiographic restenosis and consequent target-vessel revascularization, but has not affected myocardial infarction or mortality.27, 28 The ARTS-II registry enrolled patients with multivessel CAD undergoing coronary intervention with the sirolimus drug-eluting stent and compared them with patients in the ARTS-1 randomized trial.29 Rates of revascularization and major adverse cardiac and cerebrovascular events were lower for sirolimus-stent patients than for patients randomly assigned to bare-metal stents in ARTS-1. Cardiac and cerebrovascular event rates in sirolimus-stent patients were similar to those for the CABG control patients from ARTS-I, but revascularization rates were higher. The SYNTAX trial reports the first large randomized trial of drug-eluting stents versus CABG in patients with multivessel CAD.

Observational data provide stronger support for CABG than for PCI (coronary angioplasty or stent placement) in patients with multivessel CAD. In reported outcomes for >60,000 patients with multivessel CAD treated using bare-metal stents or CABG,30 at 3 years of follow-up, the repeated revascularization rate (27.3% vs 4.6% for subsequent PCI and 7.8% vs 0.3% for subsequent CABG) was greater in patients initially treated with PCI. Furthermore, both crude and adjusted 3-year mortality rates were lower in patients treated with CABG. In an analysis of patients with multivessel disease treated using drug-eluting stents or CABG,31 patients with 2- and 3-vessel CAD treated using CABG had lower revascularization, mortality, and myocardial infarction rates than patients treated using drug-eluting stents. Although intriguing, these results must be interpreted with caution because risk adjustment cannot fully account for selection bias and residual confounding.

What Should Clinicians and Researchers Do? 

Data comparing coronary revascularization with medical therapy in patients with CKD are limited by their retrospective observational nature; studies have not always compared revascularization with optimal medical therapy, including use of β-blockers. Observational studies suggest lower mortality in patients with CKD (including dialysis patients) who do versus do not undergo coronary revascularization.32, 33 A post hoc analysis of patients with CKD in the FRISC II (Fast Revascularization in Instability in Coronary Disease) trial indicated superior outcomes with an early invasive strategy in patients with acute coronary syndrome compared with conservative management,34 an approach supported by a recent systematic review.35 A post hoc analysis of patients with CKD enrolled in the ARTS trial36 found similar outcomes for CABG and multivessel PCI with bare-metal stents for death, myocardial infarction, or stroke. Unsurprisingly, CABG was associated with less repeated revascularization. Dialysis patient survival after CABG was better than after PCI with bare-metal stents.37 However, 2-year mortality was high at 44% (vs 52% for PCI). Preliminary US Renal Data System data38 suggest that drug-eluting stents provide the best 1-year survival for dialysis patients; however, unadjusted long-term survival is better after CABG and increases with the number of arteries bypassed and use of internal mammary graft conduit. Because of higher periprocedure surgical mortality, short-term survival for dialysis patients after PCI is better. However, for dialysis patients with ≥4 vessels bypassed, including the use of internal mammary grafting, survival >6 months after CABG is superior to survival with drug-eluting stents. In elderly non–dialysis-dependent patients with CKD, preliminary data suggest better 2-year survival with drug-eluting stents (with an 18% decrease in mortality compared with CABG).39

In the general population, a proven advantage of drug-eluting over bare-metal stents is lower in-stent restenosis incidence. In dialysis patients, reliance on clinical surrogates (eg, chest pain) for detecting restenosis can be problematic and lead to underestimating true incidence. The most complete angiographic studies of drug-eluting stents documented a 22%-31% angiographically detected incidence of restenosis with drug-eluting stents and 24%-40% with non–drug-eluting stents.40, 41, 42 Dialysis patients may represent the extreme case for testing (and possibly rejecting) the long-term “noninferiority” of PCI versus CABG. However, patients and researchers conducting trials may have different conceptions of “noninferior” outcomes, particularly regarding end points of interest, such as stroke, which might be seen as a more undesirable outcome than repeated revascularization.

Determining the ideal coronary revascularization strategy in patients with CKD (including end-stage renal disease) is a dilemma. Physicians must extrapolate data from contemporary clinical trials of patients with preserved kidney function and individualize decisions based on patient preference and comorbidity. The SYNTAX trial, although not restricted to patients with CKD, enrolled subjects with multivessel CAD consecutively to represent a wide range of comorbid conditions. Drug-eluting stents and modern cardiothoracic surgical techniques (off-pump CABG and use of multiple arterial conduits) represent technologic advances over prior coronary revascularization trials.

Several lessons can be gleaned from the SYNTAX trial. First, within the first year of coronary revascularization, there is no significant difference in all-cause mortality or myocardial infarction in patients with multivessel CAD who undergo PCI versus CABG. Whether the malignant nature of CAD in patients with CKD translates into a higher long-term incidence of myocardial infarction or mortality is unclear. Second, despite the introduction of drug-eluting stents, risk of repeated coronary revascularization remains higher in patients who initially underwent PCI. Although the repeated revascularization rate of 13.5% accompanying PCI may be unacceptable to some physicians and patients, others would prefer this less invasive strategy. Third, the risk of stroke is significantly decreased by a PCI strategy, possibly reflecting a more advanced form of atherosclerosis in patients with multivessel CAD in whom additional risk factors (aortic calcification, carotid artery disease, and atrial fibrillation) coexist. This may be a more important consideration in patients with CKD, in whom the hazard of stroke may be magnified by more advanced atherosclerosis and additional comorbidity. Finally, the SYNTAX score, an anatomic estimate of CAD severity, provides a pragmatic mechanism to identify high-risk patients who clearly gain greater benefit by undergoing CABG. Coronary artery calcification, an element in the SYNTAX calculation, is more prevalent in patients with CKD, particularly those with end-stage renal disease. The SYNTAX score's ability to define coronary calcification severity provides an indirect means of identifying the CKD risk factor. Additional CAD features frequently identified in patients with CKD, such as native coronary artery occlusion, tortuosity of vessels, diffuse disease pattern, small-caliber vessels, and bifurcation disease, also are incorporated into the SYNTAX score. The SYNTAX trial provides strong evidence that for patients with advanced multivessel CAD (reflected by a high SYNTAX score), CABG is the preferred method of coronary revascularization. In contemporary practice, these patients frequently are evaluated by a cardiologist and a PCI strategy is undertaken without a multidisciplinary approach involving a cardiothoracic surgeon. A better option, as shown by the SYNTAX trial, would be a staged approach for patients with multivessel CAD in which both PCI and CABG are deferred to a later time. The SYNTAX trial teaches the interventional cardiologist to consider the final lines of Robert Frost's poem The Road Not Taken: “Two roads diverged in a wood, and I—I took the one less traveled by, and that has made all the difference.”43 (Another option might be to stay at home—but that is grist for another editorial.)